Elon Musk's company xAI lost out on a massive government contract after its Grok chatbot went full Nazi last month and started calling itself "MechaHitler" while spewing racist rants, Wired reports.
It's a striking revelation, if confirmed. The Trump administration has shown little restraint in its whirlwind push to stuff AI into federal agencies. It's even used AI to help carry out its purge of DEI and so-called "woke" content in government, including deleting a page dedicated to the Black baseball legend Jackie Robinson, and ordering NASA to remove references specifically targeting women.
Backing down from the Grok deal would be a rare example of the administration — which aims to ban woke AI — apparently deeming one of its AI efforts going beyond the pale.
According to Wired, xAI was set to join a partnership that the General Services Administration, the agency that handles government technology, made with three other leading AI companies to give federal employees access to their AI tools for just $1. That partnership, which ended up including OpenAI, Google, and Anthropic, was announced last week.
The idea was first put forth by OpenAI CEO Sam Altman in May, and in June, GSA leadership met with xAI "to see what opportunities may exist for automation and streamlining," per an email obtained by Wired.
The leadership came out of the meeting enamored by Musk's chatbot, baffling their subordinates.
"We kept saying 'Are you sure?' And they were like 'No we gotta have Grok,'" one employee involved in the talks told Wired.
Soon enough, xAI was added to the GSA Multiple Award Schedule, the agency's long-term contracting program, according to Wired, which would've allowed federal agencies to buy access to Grok through an intermediary.
The deal, it seemed, was pretty much sealed until the "MechaHitler" incident happened. Beyond using this outrageously offensive sobriquet, Grok also praised the Nazi leader, called for a second Holocaust, and altogether invented new ways of disparaging Jewish people. xAI was forced to release a statement and delete some of Grok's posts, but these events went completely unacknowledged by Musk.
At first, this didn't appear to dampen leadership's enthusiasm for the bot. And again, their underlings were left stupefied.
"The week after Grok went MechaHitler, [GSA leadership] was like 'Where are we on Grok?'" the same employee told Wired. "We were like, 'Do you not read a newspaper?'"
Right before the GSA was supposed to announce its partnership with xAI, staff were told to remove Grok from the contract offering. The official reason is unclear, but two employees told Wired they believe it's because of Grok's "MechaHitler" meltdown.
The warning signs were there long before that episode, however. Grok's entire raison d'être was to be an "anti-woke" alternative to mainstream AI, and Musk has frequently declared his intent to tinker with the chatbot whenever he feels it gives responses that don't align with his extreme beliefs.
In May, Grok began to randomly ramble about claims of "white genocide" in South Africa under completely unrelated tweets. Musk, a white South African, is a proponent of the racist conspiracy theory.
Big picture-wise, getting turned down by an openly "anti-woke" administration is emblematic of how Musk's right-wing pivot can alienate fans, and more importantly, customers.
His other business venture, SpaceX, enjoys billions of dollars in government contracts; the door may have just been shut on xAI one day, enjoying the same.
Share This Article